THE ROBUSTNESS CRITERIA

In order to achieve the best quality of a process or a product, the design step of the process or product is a very important one. Most of the quality needed later can be built into the product at the design stage. Here, the best quality is defined as obtaining or approaching as close as possible those characteristics which are desired. Moreover, one wishes to keep close to these properties when variations in process conditions appear, or when a product deteriorates through use or ageing, or when external (environmental or noise) factors affect the product. 

To obtain this kind of quality, research has to be done on all the factors influencing the product. So knowledge is gathered about all the factors and their interaction(s) influencing the quality characteristic(s). This 

When limiting ourselves to one quality characteristic (y), with preferred value T, a common way to define a measure of quality is the quadratic loss function (as used by Taguchi [13]), which is defined as L(y) = k(y-f) where A is a constant, coupling the loss (y-r) to, for example, an economic quantity. 

The value of y is affected by the product design factors (factors which can be selected and maintained by the engineer) and by the environmental (noise) factors (factors which are beyond the control of the engineer). 

A low (f might be preferable, at the cost of some bias, if the attended loss is lower than a situation with no bias and larger cf. 

---

In order to assess the validity of the robustness criteria, which were derived in the previous section, an algorithm was developed for the validation of the criteria. This algorithm existed of two parts, the results of which were compared and consequently used to assess the robustness algorithm. 

---